Amy Webb has been briefing Fortune 500 boards for nearly two decades. Her annual tech reports are the ones executives actually read — concise, system-level, with enough specificity to survive a boardroom. This year she scrapped the classic “trends” format entirely.
The Convergence Outlook 2026 is a 318-page strategic playbook, and four of its ten named convergences touch biology directly. Two — Living Intelligence and Programmable Biology — sit squarely at the intersection of AI and life sciences. That’s not a coincidence. It’s a signal about where serious capital and serious attention are flowing.
The blunt line buried in the report’s leadership section: “Agentic biology is now real.” Not a forecast. Not a scenario. A present-tense declaration. Here’s what that means for everyone building at the bio-compute frontier.
Drug Discovery Has Become a Factory — Not a Breakthrough Machine
The real competitive edge in drug discovery is no longer the smartest foundation model or the most decorated biology team. It’s who can run repeatable, automated design-build-test-learn loops at high throughput — consistently, at scale, without heroic one-off insights.
Webb’s framing aligns with what we’re already watching in practice. Insilico Medicine’s deal structure with Eli Lilly — and OpenAI’s integration work with autonomous wet labs like Ginkgo Bioworks — both reflect the same architectural bet: treat biology as infrastructure that can be engineered end-to-end, not a black box that occasionally yields breakthroughs.
Calling this “faster R&D” misses the point entirely. The throughput advantage compounds. Each loop tightens the model. Each tightened model reduces the cost of the next physical validation. At scale, this is industrial-grade reprogramming of living systems — not a faster version of what pharma was doing before.
The teams that understand this are already building closed-loop pipelines where computation and physical automation are fused from day one, not bolted together after the fact.
Virtual Cells Graduated From Academic Posters to Boardroom Decks
One of the more revealing moves in the Convergence Outlook is the dedicated real estate it gives to computational cell simulation. Webb explicitly names Geneformer, TranscriptFormer, Xaira’s X-Cell model, and NVIDIA’s Virtual Cell Challenge — not as curiosities, but as strategic assets under active evaluation.
A year ago, digital twins of individual cells were primarily conference-circuit material. Now they’re appearing in strategy documents handed to global corporate leaders. That transition — from academic proof-of-concept to boardroom agenda item — is the actual news here.
Xaira’s published results reinforce a nuance Webb’s report touches on but doesn’t spell out: data diversity often beats raw model size. Their biggest performance jumps came from training across 152,000+ unique experimental conditions, not from scaling parameter count. That’s a direct challenge to the “just build bigger models” playbook.
What this means practically: the bottleneck in virtual cell utility isn’t compute. It’s data quality, experimental diversity, and the engineering required to translate simulation outputs into reliable wet-lab predictions.
“Agentic Biology” Is an Infrastructure Claim, Not a Lab Claim
The phrase that’s been circulating since journalist Peter Ottsjö’s thread broke the report down publicly: “Agentic biology is now real.” Webb uses it to describe what leaders are most systematically underestimating.
Agentic biology refers to biological systems — and bio-AI hybrids — that can sense, decide, and act in closed loops, operating as running infrastructure rather than discrete experimental inputs. The shift is categorical, not incremental.
Webb’s analogy to cloud computing is apt. At some point in the 2010s, “just servers” stopped being the frame for AWS and became “the backbone of everything else.” Biological systems — organoids, engineered cell circuits, wetware-AI hybrids — are crossing an equivalent threshold. The category error is still calling this “biotech R&D.”
At BioComputer we’ve been tracking the early signals of this shift: Cortical Labs’ CL1 chip, FinalSpark’s Neuroplatform, logic-gated CAR-T therapies that make treatment decisions at the cellular level. These aren’t isolated demos. They’re proof that the sensing-deciding-acting loop is achievable in biological substrate — and that it can be engineered with intent.
The Full Stack Is the Moat — Not the Model
Webb’s two convergences — Living Intelligence and Programmable Biology — map to what we’d call the top and bottom of the biocomputer stack. Programmable Biology turns DNA, RNA, and proteins into editable infrastructure. Living Intelligence merges biological and machine intelligence into systems that adapt in real time.
The teams that will define the next decade aren’t the ones with the best single-layer innovation. They’re the ones solving the full vertical:
- Higher-quality, more diverse biological data generation
- Models trained on that diversity — not just scaled on volume
- Fast, automated physical validation pipelines
- Tight closed loops where each experiment informs the next
Protein design is still largely artisanal in most pipelines today. The throughput advantage belongs to whoever industrializes the feedback loop first — and keeps it running.
The Report Is Already Priced In, But the Work Isn’t
Amy Webb naming these forces in a 318-page document read by Fortune 500 boards changes something. Not the science — the science is already moving. What changes is the mandate. Boards that previously needed a champion to greenlight bio-AI investment now have an analyst they already trust telling them it’s an economic imperative.
That’s a lagging signal, not a leading one. The leading signal was the actual work: Insilico’s Phase II results, Xaira’s data diversity findings, Cortical Labs’ roadmap. Webb synthesizes and legitimizes. The builders are already three steps ahead.
The real question isn’t whether agentic biology is real. It’s who controls the infrastructure layer when it becomes as invisible and load-bearing as cloud compute.
References
- Webb, A. (2026). Convergence Outlook 2026. Future Today Strategic Group. https://ftsg.com/convergence
- Ottsjö, P. (2026). Thread on Amy Webb’s Convergence Outlook 2026. X / Twitter. https://twitter.com/peterottsjo
- Xaira Therapeutics. (2025). X-Cell model and experimental condition diversity results. Xaira Research. https://xaira.com
- NVIDIA. (2025). Virtual Cell Challenge. NVIDIA Research. https://research.nvidia.com/virtual-cell
- Insilico Medicine. (2025). Lilly partnership and pipeline updates. Insilico Medicine. https://insilico.com
Related: What Is a Biocomputer in 2026? · FinalSpark Neuroplatform: The First Commercial Wetware Cloud · Cortical Labs CL1: Biology Meets Silicon
Feature image: AI-generated using Grok.